Discrimination Models and Tools for Published and Deposited Data

The project will investigate mathematical model of decision processes in terms of choice proportions and speed for a wide range of psychological areas, including: perceptual discrimination, memory, attitudes and behavioural choice, cancer detection from images, terrorist identification by experts.

The models evaluated will be random walk models with theoretical parameters comprising: information accrual rate (dependent on task difficulty); barrier locations (dependent on anount of time person decides to spend to reach each response) and non-decision time.

Raw input will comprise person identification, stimulus, response, and one or more predictive factors, such as difficulty and/or  time pressure.
The project will produce an Open Source software tool in the language R to generate information accrual parameters for each participant in each condition as input to further analyses to evaluate the effect lof predictors on the information accrual parameters.

The tool will enable
* meta-analyses of which predictors effect which information accrual parameters in a very wide range of domains, theoretical and practical.
* identifiaction of which class of  information accrual model (linear ballistic accurmulator or decision diffusion model) best fit data in each domain.
* applied and theoretical researchers, to SIMPLY evaluate whether predictorsare predominantly affecting task proficiency or speed pressure.

The tool will enable consolidatin of existing knowledge and generation of new research forimportant theoretical and practical domains.

Layman's description
People often have to make many similar choices between two alternatives in conditions of uncertainty For example, a raiographer may have to examine 100s of slides for abnormalities that might idicate cancer; or a security guard may have to scan many people to decide who to search for bombs. These are binary decisions. Or we may want to evaluate the limits of human hearing by presnting a soft sound or nothing on many trials.

Perforamce is usually assessed bye either speed [for easier tasks] or accuracy [for harder tasks].

This project evaluates THEORETICAL models that asume people collect information of quality that depends on task difficulty, V, [e.g. how much do cancer cells 'stand out'] and time peopel CHOOSE to go on collecting information], A.

This project aims top provide a tool that estimates, V and A. this enables:
* identifying individuas who are good at task, high V.
* identifying conditions that make task easier, and by how much (difference in V )e.g. better lighting.
* the effects of speed prressure that will decrease time spent on task, A, but at what cost in accuracy?

The tool will enable:
* results form many previous studies to be combined to consolidate and illuminate existing knowledge
* researchers with any new project to eassily identify effects or personal ability, task difficulty and time pressure.